Where new members quietly fall off in their first 90 days

A member who cancels in month seven usually decided to leave much earlier than that. Year-one retention is mostly won or lost in the first 90 days, the stretch where a new member either builds a habit or starts drifting. The problem is that most onboarding programs run on a calendar that has no idea which one is happening.

The good news: the moment a new member starts drifting is visible in the reservation data you already have, weeks before it shows up as a cancellation.

The problem: onboarding runs on a calendar, churn runs on behavior

Most onboarding programs are date-based: a welcome email on day 1, a check-in on day 30, a how's-it-going on day 60, sent on schedule regardless of what the member has actually done in between. It feels organized, and it treats every new member as identical.

But whether someone sticks isn't decided by the calendar. It's decided by behavior:

• Did they come back for a second class within the first week?
• Did they settle into a weekly rhythm?
• Did they rebook when their intro pack ran out?

Two members who both look like new, 45 days in, can be on completely opposite paths: one building a routine, one already gone in every way except the billing record.

Why it's hard to see in your day-to-day reports

Booking platforms are built around bookings, memberships, and payments, not around the shape of a new member's early behavior. Their default reports tell you who is active and who cancelled. A new member who is quietly drifting looks exactly like a healthy one: still active, still inside their intro period, right up until the day they cancel.

The signal that predicts whether they'll stay lives in reservation history, not in the membership status field. And it's rarely a round number. It's usually something specific and unglamorous, like missed two of their first four scheduled weeks, or never booked a third class after the intro pack, the kind of pattern you only see when you look at early behavior as a trend instead of a snapshot.

How to find it in your data

• Pull first-90-day reservation history for members grouped by outcome: those who stayed past a year versus those who cancelled inside it. This is the raw booking record you already have for every member.
• Compare their early behavior week by week: visits in week 1, whether they rebooked within 7 days of the first class, their longest gap between visits, whether they let a package lapse without rebooking.
• Find the point where the two groups diverge. That's your at-risk milestone, and it's often earlier than owners expect, closer to week 2 or 3 than day 60. This works whether your data lives in Mariana Tek, MindBody, or PushPress, since all three store the same underlying reservation history.
• Turn it into a weekly flag. Any new member who crosses that behavioral line gets surfaced while there's still time to change the outcome, instead of after the intro period quietly expires.

The exact threshold matters less than the direction. A new member whose early pattern has visibly stalled is worth a touch regardless of the precise cutoff you land on.

What to do about it

Detection only helps if it changes the timing of something you already do. The point of catching a stall at week 2 is that you still have real options that don't exist once the intro period has lapsed:

• Match the touch to the gap: no second booking yet gets a message with three specific open class times, not a generic come back soon. Package ran out gets a rebooking prompt before the lapse, not a win-back weeks later. A long gap gets a personal check-in from the instructor they actually know.
• Trigger on the behavior, not the date: send the touch when the member crosses the at-risk line, not on a fixed day that may land after they've already drifted.
• Close the loop: track whether flagged members who got a touch return within the next two weeks, compared to those who didn't. That tells you if the intervention is working, and which gap deserves the most attention.

Most owners are still guessing where their new members fall off, or relying on the memory of a front desk that knows a few regulars by name. The data to answer it precisely is already sitting in your booking platform.